Project realized during my second year at the ENSAE in partnership with the CREST (Centerfor Research in Economics and Statistics).
The Goal was to used unsupervised learning to Identify and Classify driving style with GPS data.
The data comes from a Kaggle Challenge proprosed by Axa insurance.
This repository presents only a part of the entire project. We used Heatmaps of velocity and acceleration to classify the drivers thanks to image unsupervised classification with convolutionnal neuron network.